Diese Präsentation wurde erfolgreich gemeldet.
Wir verwenden Ihre LinkedIn Profilangaben und Informationen zu Ihren Aktivitäten, um Anzeigen zu personalisieren und Ihnen relevantere Inhalte anzuzeigen. Sie können Ihre Anzeigeneinstellungen jederzeit ändern.
Human-in-the-Loop Fine-Grained Visual    Categorization using Visipedia                  or   "The Revolution will be Cura...
2
Birds-200 Dataset6033 images over 200 bird species
4
MTurker Label Certainty                          5
Visual 20 Questions• “Computer Vision” module = Vedaldi’s VLFeat• VQ Geometric Blur, color/gray SIFT spatial pyramid• Mult...
7
General Observations• User Responses are Stochastic• Computer Vision Reduces Manual Labor• User Responses Drive Up Perform...
w/o Computer Vision• User Responses are Stochastic                              9
w/ Computer Vision• Computer Vision Reduces Manual Labor                                 10
w/ Computer Vision (cont’d)• User Responses Drive Up Performance                                11
• Computer Vision Improves Overall Performance• Different Questions are Asked w/ and w/o  Computer Vision
• Recognition is not Always Successful
Indigo Bunting   Blue Grosbeak
Fcv the revolution will be curated: human in the loop fine grained visual categorization using visipedia belongie
Fcv the revolution will be curated: human in the loop fine grained visual categorization using visipedia belongie
Nächste SlideShare
Wird geladen in …5
×

Fcv the revolution will be curated: human in the loop fine grained visual categorization using visipedia belongie

  • Loggen Sie sich ein, um Kommentare anzuzeigen.

Fcv the revolution will be curated: human in the loop fine grained visual categorization using visipedia belongie

  1. 1. Human-in-the-Loop Fine-Grained Visual Categorization using Visipedia or "The Revolution will be Curated" Serge Belongie UC San Diego Steve Branson Catherine Wah Peter Welinder Boris Babenko Pietro Perona Florian Schroff
  2. 2. 2
  3. 3. Birds-200 Dataset6033 images over 200 bird species
  4. 4. 4
  5. 5. MTurker Label Certainty 5
  6. 6. Visual 20 Questions• “Computer Vision” module = Vedaldi’s VLFeat• VQ Geometric Blur, color/gray SIFT spatial pyramid• Multiple Kernel Learning• Per-Class 1-vs-All SVM• 15 training examples per bird species• Choose question to maximize expected Information Gain 6
  7. 7. 7
  8. 8. General Observations• User Responses are Stochastic• Computer Vision Reduces Manual Labor• User Responses Drive Up Performance• Computer Vision Improves Overall Performance• Different Questions are Asked w/ and w/o Computer Vision• Recognition is not Always Successful 8
  9. 9. w/o Computer Vision• User Responses are Stochastic 9
  10. 10. w/ Computer Vision• Computer Vision Reduces Manual Labor 10
  11. 11. w/ Computer Vision (cont’d)• User Responses Drive Up Performance 11
  12. 12. • Computer Vision Improves Overall Performance• Different Questions are Asked w/ and w/o Computer Vision
  13. 13. • Recognition is not Always Successful
  14. 14. Indigo Bunting Blue Grosbeak

×